How power constraints could limit the AI boom
How power constraints could limit the AI boom
About this video
Key points:
Higher rates are unlikely to derail hyperscaler AI spending, with major technology companies expected to remain resilient over the next 6–12 months.
Smaller AI infrastructure players face greater funding pressure as the cost of capital rises.
Around $1.4 trillion of AI infrastructure spending is expected from major players over the next two years, supporting demand across the hardware supply chain.
Power availability and potential double-ordering are key risks, while component supply constraints could persist as AI models become more advanced.
Higher interest rates are unlikely to significantly disrupt spending by major technology companies on artificial intelligence infrastructure, although smaller players could face greater pressure from rising funding costs, according to Circular Technology’s Brad Gastwirth.
Gastwirth says the Federal Reserve’s 25-basis-point rate hike was largely priced into markets, with investors now focused on the prospect of rates remaining higher for longer. He expects hyperscalers to remain resilient over the next six to 12 months, while second- and third-tier companies could be more exposed as the cost of capital rises. The AI infrastructure buildout is also creating growing demand for power, with backup generation becoming an increasingly important constraint for data centres. Gastwirth points to Amazon’s $8 billion agreement with Generac as an example of the scale of investment flowing into the sector.
Gastwirth says AI infrastructure spending is likely to remain elevated, with around $1.4 trillion expected to be spent by a handful of major players over the next two years. He expects more companies to raise capital as they fund expansion, particularly if interest rates continue to rise. Meanwhile, he says demand across the AI hardware supply chain remains strong, although investors should watch for double or triple ordering that could exaggerate underlying demand. Gastwirth expects supply constraints across some components to persist as the industry continues its push towards increasingly advanced AI models.